7 # Multi-level database module for storing hash trees, arrays and simple
8 # key/value pairs into FTP-able, cross-platform binary database files.
10 # Type `perldoc DBM::Deep` for complete documentation.
14 # tie %db, 'DBM::Deep', 'my_database.db'; # standard tie() method
16 # my $db = new DBM::Deep( 'my_database.db' ); # preferred OO method
18 # $db->{my_scalar} = 'hello world';
19 # $db->{my_hash} = { larry => 'genius', hashes => 'fast' };
20 # $db->{my_array} = [ 1, 2, 3, time() ];
21 # $db->{my_complex} = [ 'hello', { perl => 'rules' }, 42, 99 ];
22 # push @{$db->{my_array}}, 'another value';
23 # my @key_list = keys %{$db->{my_hash}};
24 # print "This module " . $db->{my_complex}->[1]->{perl} . "!\n";
27 # (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
28 # This program is free software; you can redistribute it and/or
29 # modify it under the same terms as Perl itself.
34 use Fcntl qw( :DEFAULT :flock :seek );
38 use DBM::Deep::Engine;
40 use vars qw( $VERSION );
41 $VERSION = q(0.99_01);
44 # Setup constants for users to pass to new()
46 sub TYPE_HASH () { DBM::Deep::Engine::SIG_HASH }
47 sub TYPE_ARRAY () { DBM::Deep::Engine::SIG_ARRAY }
48 sub TYPE_SCALAR () { DBM::Deep::Engine::SIG_SCALAR }
56 $proto->_throw_error( "Odd number of parameters to " . (caller(1))[2] );
61 unless ( eval { local $SIG{'__DIE__'}; %{$_[0]} || 1 } ) {
62 $proto->_throw_error( "Not a hashref in args to " . (caller(1))[2] );
67 $args = { file => shift };
75 # Class constructor method for Perl OO interface.
76 # Calls tie() and returns blessed reference to tied hash or array,
77 # providing a hybrid OO/tie interface.
80 my $args = $class->_get_args( @_ );
83 # Check if we want a tied hash or array.
86 if (defined($args->{type}) && $args->{type} eq TYPE_ARRAY) {
87 $class = 'DBM::Deep::Array';
88 require DBM::Deep::Array;
89 tie @$self, $class, %$args;
92 $class = 'DBM::Deep::Hash';
93 require DBM::Deep::Hash;
94 tie %$self, $class, %$args;
97 return bless $self, $class;
102 # Setup $self and bless into this class.
107 # These are the defaults to be optionally overridden below
110 engine => DBM::Deep::Engine->new,
112 $self->{base_offset} = length( $self->{engine}->SIG_FILE );
114 foreach my $param ( keys %$self ) {
115 next unless exists $args->{$param};
116 $self->{$param} = delete $args->{$param}
119 # locking implicitly enables autoflush
120 if ($args->{locking}) { $args->{autoflush} = 1; }
122 $self->{root} = exists $args->{root}
124 : DBM::Deep::_::Root->new( $args );
126 $self->{engine}->setup_fh( $self );
133 require DBM::Deep::Hash;
134 return DBM::Deep::Hash->TIEHASH( @_ );
139 require DBM::Deep::Array;
140 return DBM::Deep::Array->TIEARRAY( @_ );
143 #XXX Unneeded now ...
149 # If db locking is set, flock() the db file. If called multiple
150 # times before unlock(), then the same number of unlocks() must
151 # be called before the lock is released.
153 my $self = shift->_get_self;
155 $type = LOCK_EX unless defined $type;
157 if (!defined($self->_fh)) { return; }
159 if ($self->_root->{locking}) {
160 if (!$self->_root->{locked}) {
161 flock($self->_fh, $type);
163 # refresh end counter in case file has changed size
164 my @stats = stat($self->_root->{file});
165 $self->_root->{end} = $stats[7];
167 # double-check file inode, in case another process
168 # has optimize()d our file while we were waiting.
169 if ($stats[1] != $self->_root->{inode}) {
170 $self->{engine}->close_fh( $self );
171 $self->{engine}->setup_fh( $self );
172 flock($self->_fh, $type); # re-lock
174 # This may not be necessary after re-opening
175 $self->_root->{end} = (stat($self->_fh))[7]; # re-end
178 $self->_root->{locked}++;
188 # If db locking is set, unlock the db file. See note in lock()
189 # regarding calling lock() multiple times.
191 my $self = shift->_get_self;
193 if (!defined($self->_fh)) { return; }
195 if ($self->_root->{locking} && $self->_root->{locked} > 0) {
196 $self->_root->{locked}--;
197 if (!$self->_root->{locked}) { flock($self->_fh, LOCK_UN); }
206 my $self = shift->_get_self;
207 my ($spot, $value) = @_;
212 elsif ( eval { local $SIG{__DIE__}; $value->isa( 'DBM::Deep' ) } ) {
213 my $type = $value->_type;
214 ${$spot} = $type eq TYPE_HASH ? {} : [];
215 $value->_copy_node( ${$spot} );
218 my $r = Scalar::Util::reftype( $value );
219 my $c = Scalar::Util::blessed( $value );
220 if ( $r eq 'ARRAY' ) {
221 ${$spot} = [ @{$value} ];
224 ${$spot} = { %{$value} };
226 ${$spot} = bless ${$spot}, $c
235 # Copy single level of keys or elements to new DB handle.
236 # Recurse for nested structures
238 my $self = shift->_get_self;
241 if ($self->_type eq TYPE_HASH) {
242 my $key = $self->first_key();
244 my $value = $self->get($key);
245 $self->_copy_value( \$db_temp->{$key}, $value );
246 $key = $self->next_key($key);
250 my $length = $self->length();
251 for (my $index = 0; $index < $length; $index++) {
252 my $value = $self->get($index);
253 $self->_copy_value( \$db_temp->[$index], $value );
262 # Recursively export into standard Perl hashes and arrays.
264 my $self = shift->_get_self;
267 if ($self->_type eq TYPE_HASH) { $temp = {}; }
268 elsif ($self->_type eq TYPE_ARRAY) { $temp = []; }
271 $self->_copy_node( $temp );
279 # Recursively import Perl hash/array structure
281 if (!ref($_[0])) { return; } # Perl calls import() on use -- ignore
283 my $self = shift->_get_self;
286 # struct is not a reference, so just import based on our type
288 if ($self->_type eq TYPE_HASH) { $struct = {@_}; }
289 elsif ($self->_type eq TYPE_ARRAY) { $struct = [@_]; }
292 my $r = Scalar::Util::reftype($struct) || '';
293 if ($r eq "HASH" && $self->_type eq TYPE_HASH) {
294 foreach my $key (keys %$struct) { $self->put($key, $struct->{$key}); }
296 elsif ($r eq "ARRAY" && $self->_type eq TYPE_ARRAY) {
297 $self->push( @$struct );
300 $self->_throw_error("Cannot import: type mismatch");
308 # Rebuild entire database into new file, then move
309 # it back on top of original.
311 my $self = shift->_get_self;
313 #XXX Need to create a new test for this
314 # if ($self->_root->{links} > 1) {
315 # $self->_throw_error("Cannot optimize: reference count is greater than 1");
318 my $db_temp = DBM::Deep->new(
319 file => $self->_root->{file} . '.tmp',
323 $self->_throw_error("Cannot optimize: failed to open temp file: $!");
327 $self->_copy_node( $db_temp );
331 # Attempt to copy user, group and permissions over to new file
333 my @stats = stat($self->_fh);
334 my $perms = $stats[2] & 07777;
337 chown( $uid, $gid, $self->_root->{file} . '.tmp' );
338 chmod( $perms, $self->_root->{file} . '.tmp' );
340 # q.v. perlport for more information on this variable
341 if ( $^O eq 'MSWin32' || $^O eq 'cygwin' ) {
343 # Potential race condition when optmizing on Win32 with locking.
344 # The Windows filesystem requires that the filehandle be closed
345 # before it is overwritten with rename(). This could be redone
349 $self->{engine}->close_fh( $self );
352 if (!rename $self->_root->{file} . '.tmp', $self->_root->{file}) {
353 unlink $self->_root->{file} . '.tmp';
355 $self->_throw_error("Optimize failed: Cannot copy temp file over original: $!");
359 $self->{engine}->close_fh( $self );
360 $self->{engine}->setup_fh( $self );
367 # Make copy of object and return
369 my $self = shift->_get_self;
371 return DBM::Deep->new(
372 type => $self->_type,
373 base_offset => $self->_base_offset,
379 my %is_legal_filter = map {
382 store_key store_value
383 fetch_key fetch_value
388 # Setup filter function for storing or fetching the key or value
390 my $self = shift->_get_self;
394 if ( $is_legal_filter{$type} ) {
395 $self->_root->{"filter_$type"} = $func;
409 # Get access to the root structure
411 my $self = $_[0]->_get_self;
412 return $self->{root};
417 # Get type of current node (TYPE_HASH or TYPE_ARRAY)
419 my $self = $_[0]->_get_self;
420 return $self->{type};
425 # Get base_offset of current node (TYPE_HASH or TYPE_ARRAY)
427 my $self = $_[0]->_get_self;
428 return $self->{base_offset};
433 # Get access to the raw fh
435 my $self = $_[0]->_get_self;
436 return $self->_root->{fh};
444 die "DBM::Deep: $_[1]\n";
449 (O_WRONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
454 # (O_RDONLY | O_RDWR) & fcntl( $fh, F_GETFL, my $slush = 0);
459 # Store single hash key/value or array element in database.
461 my $self = shift->_get_self;
462 my ($key, $value) = @_;
464 unless ( _is_writable( $self->_fh ) ) {
465 $self->_throw_error( 'Cannot write to a readonly filehandle' );
469 # Request exclusive lock for writing
471 $self->lock( LOCK_EX );
473 my $md5 = $self->{engine}{digest}->($key);
475 my $tag = $self->{engine}->find_bucket_list( $self, $md5, { create => 1 } );
477 # User may be storing a hash, in which case we do not want it run
478 # through the filtering system
479 if ( !ref($value) && $self->_root->{filter_store_value} ) {
480 $value = $self->_root->{filter_store_value}->( $value );
484 # Add key/value to bucket list
486 my $result = $self->{engine}->add_bucket( $self, $tag, $md5, $key, $value );
495 # Fetch single value or element given plain key or array index
497 my $self = shift->_get_self;
500 my $md5 = $self->{engine}{digest}->($key);
503 # Request shared lock for reading
505 $self->lock( LOCK_SH );
507 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
514 # Get value from bucket list
516 my $result = $self->{engine}->get_bucket_value( $self, $tag, $md5 );
520 # Filters only apply to scalar values, so the ref check is making
521 # sure the fetched bucket is a scalar, not a child hash or array.
522 return ($result && !ref($result) && $self->_root->{filter_fetch_value})
523 ? $self->_root->{filter_fetch_value}->($result)
529 # Delete single key/value pair or element given plain key or array index
531 my $self = $_[0]->_get_self;
534 unless ( _is_writable( $self->_fh ) ) {
535 $self->_throw_error( 'Cannot write to a readonly filehandle' );
539 # Request exclusive lock for writing
541 $self->lock( LOCK_EX );
543 my $md5 = $self->{engine}{digest}->($key);
545 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
554 my $value = $self->{engine}->get_bucket_value($self, $tag, $md5 );
556 if (defined $value && !ref($value) && $self->_root->{filter_fetch_value}) {
557 $value = $self->_root->{filter_fetch_value}->($value);
560 my $result = $self->{engine}->delete_bucket( $self, $tag, $md5 );
563 # If this object is an array and the key deleted was on the end of the stack,
564 # decrement the length variable.
574 # Check if a single key or element exists given plain key or array index
576 my $self = $_[0]->_get_self;
579 my $md5 = $self->{engine}{digest}->($key);
582 # Request shared lock for reading
584 $self->lock( LOCK_SH );
586 my $tag = $self->{engine}->find_bucket_list( $self, $md5 );
591 # For some reason, the built-in exists() function returns '' for false
597 # Check if bucket exists and return 1 or ''
599 my $result = $self->{engine}->bucket_exists( $self, $tag, $md5 ) || '';
608 # Clear all keys from hash, or all elements from array.
610 my $self = $_[0]->_get_self;
612 unless ( _is_writable( $self->_fh ) ) {
613 $self->_throw_error( 'Cannot write to a readonly filehandle' );
617 # Request exclusive lock for writing
619 $self->lock( LOCK_EX );
623 seek($fh, $self->_base_offset + $self->_root->{file_offset}, SEEK_SET);
629 $self->{engine}->create_tag($self, $self->_base_offset, $self->_type, chr(0) x $self->{engine}{index_size});
637 # Public method aliases
639 sub put { (shift)->STORE( @_ ) }
640 sub store { (shift)->STORE( @_ ) }
641 sub get { (shift)->FETCH( @_ ) }
642 sub fetch { (shift)->FETCH( @_ ) }
643 sub delete { (shift)->DELETE( @_ ) }
644 sub exists { (shift)->EXISTS( @_ ) }
645 sub clear { (shift)->CLEAR( @_ ) }
647 package DBM::Deep::_::Root;
656 #XXX It should be this in order to work with the initial create_tag(),
657 #XXX but it's not ... it works out because of the stat() in setup_fh(),
658 #XXX but that's not good.
659 end => 0, #length(DBM::Deep->SIG_FILE),
665 filter_store_key => undef,
666 filter_store_value => undef,
667 filter_fetch_key => undef,
668 filter_fetch_value => undef,
672 if ( $self->{fh} && !$self->{file_offset} ) {
673 $self->{file_offset} = tell( $self->{fh} );
683 close $self->{fh} if $self->{fh};
693 DBM::Deep - A pure perl multi-level hash/array DBM
698 my $db = DBM::Deep->new( "foo.db" );
700 $db->{key} = 'value'; # tie() style
703 $db->put('key' => 'value'); # OO style
704 print $db->get('key');
706 # true multi-level support
707 $db->{my_complex} = [
708 'hello', { perl => 'rules' },
714 A unique flat-file database module, written in pure perl. True
715 multi-level hash/array support (unlike MLDBM, which is faked), hybrid
716 OO / tie() interface, cross-platform FTPable files, and quite fast. Can
717 handle millions of keys and unlimited hash levels without significant
718 slow-down. Written from the ground-up in pure perl -- this is NOT a
719 wrapper around a C-based DBM. Out-of-the-box compatibility with Unix,
720 Mac OS X and Windows.
722 =head1 VERSION DIFFERENCES
724 B<NOTE>: 0.99_01 and above have significant file format differences from 0.98 and
725 before. While attempts have been made to be backwards compatible, no guarantees.
729 Hopefully you are using Perl's excellent CPAN module, which will download
730 and install the module for you. If not, get the tarball, and run these
742 Construction can be done OO-style (which is the recommended way), or using
743 Perl's tie() function. Both are examined here.
745 =head2 OO CONSTRUCTION
747 The recommended way to construct a DBM::Deep object is to use the new()
748 method, which gets you a blessed, tied hash or array reference.
750 my $db = DBM::Deep->new( "foo.db" );
752 This opens a new database handle, mapped to the file "foo.db". If this
753 file does not exist, it will automatically be created. DB files are
754 opened in "r+" (read/write) mode, and the type of object returned is a
755 hash, unless otherwise specified (see L<OPTIONS> below).
757 You can pass a number of options to the constructor to specify things like
758 locking, autoflush, etc. This is done by passing an inline hash:
760 my $db = DBM::Deep->new(
766 Notice that the filename is now specified I<inside> the hash with
767 the "file" parameter, as opposed to being the sole argument to the
768 constructor. This is required if any options are specified.
769 See L<OPTIONS> below for the complete list.
773 You can also start with an array instead of a hash. For this, you must
774 specify the C<type> parameter:
776 my $db = DBM::Deep->new(
778 type => DBM::Deep->TYPE_ARRAY
781 B<Note:> Specifing the C<type> parameter only takes effect when beginning
782 a new DB file. If you create a DBM::Deep object with an existing file, the
783 C<type> will be loaded from the file header, and an error will be thrown if
784 the wrong type is passed in.
786 =head2 TIE CONSTRUCTION
788 Alternately, you can create a DBM::Deep handle by using Perl's built-in
789 tie() function. The object returned from tie() can be used to call methods,
790 such as lock() and unlock(), but cannot be used to assign to the DBM::Deep
791 file (as expected with most tie'd objects).
794 my $db = tie %hash, "DBM::Deep", "foo.db";
797 my $db = tie @array, "DBM::Deep", "bar.db";
799 As with the OO constructor, you can replace the DB filename parameter with
800 a hash containing one or more options (see L<OPTIONS> just below for the
803 tie %hash, "DBM::Deep", {
811 There are a number of options that can be passed in when constructing your
812 DBM::Deep objects. These apply to both the OO- and tie- based approaches.
818 Filename of the DB file to link the handle to. You can pass a full absolute
819 filesystem path, partial path, or a plain filename if the file is in the
820 current working directory. This is a required parameter (though q.v. fh).
824 If you want, you can pass in the fh instead of the file. This is most useful for doing
827 my $db = DBM::Deep->new( { fh => \*DATA } );
829 You are responsible for making sure that the fh has been opened appropriately for your
830 needs. If you open it read-only and attempt to write, an exception will be thrown. If you
831 open it write-only or append-only, an exception will be thrown immediately as DBM::Deep
832 needs to read from the fh.
836 This is the offset within the file that the DBM::Deep db starts. Most of the time, you will
837 not need to set this. However, it's there if you want it.
839 If you pass in fh and do not set this, it will be set appropriately.
843 This parameter specifies what type of object to create, a hash or array. Use
844 one of these two constants: C<DBM::Deep-E<gt>TYPE_HASH> or C<DBM::Deep-E<gt>TYPE_ARRAY>.
845 This only takes effect when beginning a new file. This is an optional
846 parameter, and defaults to C<DBM::Deep-E<gt>TYPE_HASH>.
850 Specifies whether locking is to be enabled. DBM::Deep uses Perl's Fnctl flock()
851 function to lock the database in exclusive mode for writes, and shared mode for
852 reads. Pass any true value to enable. This affects the base DB handle I<and
853 any child hashes or arrays> that use the same DB file. This is an optional
854 parameter, and defaults to 0 (disabled). See L<LOCKING> below for more.
858 Specifies whether autoflush is to be enabled on the underlying filehandle.
859 This obviously slows down write operations, but is required if you may have
860 multiple processes accessing the same DB file (also consider enable I<locking>).
861 Pass any true value to enable. This is an optional parameter, and defaults to 0
866 If I<autobless> mode is enabled, DBM::Deep will preserve blessed hashes, and
867 restore them when fetched. This is an B<experimental> feature, and does have
868 side-effects. Basically, when hashes are re-blessed into their original
869 classes, they are no longer blessed into the DBM::Deep class! So you won't be
870 able to call any DBM::Deep methods on them. You have been warned.
871 This is an optional parameter, and defaults to 0 (disabled).
875 See L<FILTERS> below.
881 With DBM::Deep you can access your databases using Perl's standard hash/array
882 syntax. Because all DBM::Deep objects are I<tied> to hashes or arrays, you can
883 treat them as such. DBM::Deep will intercept all reads/writes and direct them
884 to the right place -- the DB file. This has nothing to do with the
885 L<TIE CONSTRUCTION> section above. This simply tells you how to use DBM::Deep
886 using regular hashes and arrays, rather than calling functions like C<get()>
887 and C<put()> (although those work too). It is entirely up to you how to want
888 to access your databases.
892 You can treat any DBM::Deep object like a normal Perl hash reference. Add keys,
893 or even nested hashes (or arrays) using standard Perl syntax:
895 my $db = DBM::Deep->new( "foo.db" );
897 $db->{mykey} = "myvalue";
899 $db->{myhash}->{subkey} = "subvalue";
901 print $db->{myhash}->{subkey} . "\n";
903 You can even step through hash keys using the normal Perl C<keys()> function:
905 foreach my $key (keys %$db) {
906 print "$key: " . $db->{$key} . "\n";
909 Remember that Perl's C<keys()> function extracts I<every> key from the hash and
910 pushes them onto an array, all before the loop even begins. If you have an
911 extra large hash, this may exhaust Perl's memory. Instead, consider using
912 Perl's C<each()> function, which pulls keys/values one at a time, using very
915 while (my ($key, $value) = each %$db) {
916 print "$key: $value\n";
919 Please note that when using C<each()>, you should always pass a direct
920 hash reference, not a lookup. Meaning, you should B<never> do this:
923 while (my ($key, $value) = each %{$db->{foo}}) { # BAD
925 This causes an infinite loop, because for each iteration, Perl is calling
926 FETCH() on the $db handle, resulting in a "new" hash for foo every time, so
927 it effectively keeps returning the first key over and over again. Instead,
928 assign a temporary variable to C<$db->{foo}>, then pass that to each().
932 As with hashes, you can treat any DBM::Deep object like a normal Perl array
933 reference. This includes inserting, removing and manipulating elements,
934 and the C<push()>, C<pop()>, C<shift()>, C<unshift()> and C<splice()> functions.
935 The object must have first been created using type C<DBM::Deep-E<gt>TYPE_ARRAY>,
936 or simply be a nested array reference inside a hash. Example:
938 my $db = DBM::Deep->new(
939 file => "foo-array.db",
940 type => DBM::Deep->TYPE_ARRAY
944 push @$db, "bar", "baz";
947 my $last_elem = pop @$db; # baz
948 my $first_elem = shift @$db; # bah
949 my $second_elem = $db->[1]; # bar
951 my $num_elements = scalar @$db;
955 In addition to the I<tie()> interface, you can also use a standard OO interface
956 to manipulate all aspects of DBM::Deep databases. Each type of object (hash or
957 array) has its own methods, but both types share the following common methods:
958 C<put()>, C<get()>, C<exists()>, C<delete()> and C<clear()>.
962 =item * new() / clone()
964 These are the constructor and copy-functions.
966 =item * put() / store()
968 Stores a new hash key/value pair, or sets an array element value. Takes two
969 arguments, the hash key or array index, and the new value. The value can be
970 a scalar, hash ref or array ref. Returns true on success, false on failure.
972 $db->put("foo", "bar"); # for hashes
973 $db->put(1, "bar"); # for arrays
975 =item * get() / fetch()
977 Fetches the value of a hash key or array element. Takes one argument: the hash
978 key or array index. Returns a scalar, hash ref or array ref, depending on the
981 my $value = $db->get("foo"); # for hashes
982 my $value = $db->get(1); # for arrays
986 Checks if a hash key or array index exists. Takes one argument: the hash key
987 or array index. Returns true if it exists, false if not.
989 if ($db->exists("foo")) { print "yay!\n"; } # for hashes
990 if ($db->exists(1)) { print "yay!\n"; } # for arrays
994 Deletes one hash key/value pair or array element. Takes one argument: the hash
995 key or array index. Returns true on success, false if not found. For arrays,
996 the remaining elements located after the deleted element are NOT moved over.
997 The deleted element is essentially just undefined, which is exactly how Perl's
998 internal arrays work. Please note that the space occupied by the deleted
999 key/value or element is B<not> reused again -- see L<UNUSED SPACE RECOVERY>
1000 below for details and workarounds.
1002 $db->delete("foo"); # for hashes
1003 $db->delete(1); # for arrays
1007 Deletes B<all> hash keys or array elements. Takes no arguments. No return
1008 value. Please note that the space occupied by the deleted keys/values or
1009 elements is B<not> reused again -- see L<UNUSED SPACE RECOVERY> below for
1010 details and workarounds.
1012 $db->clear(); # hashes or arrays
1014 =item * lock() / unlock()
1020 Recover lost disk space.
1022 =item * import() / export()
1024 Data going in and out.
1026 =item * set_digest() / set_pack() / set_filter()
1028 q.v. adjusting the interal parameters.
1034 For hashes, DBM::Deep supports all the common methods described above, and the
1035 following additional methods: C<first_key()> and C<next_key()>.
1041 Returns the "first" key in the hash. As with built-in Perl hashes, keys are
1042 fetched in an undefined order (which appears random). Takes no arguments,
1043 returns the key as a scalar value.
1045 my $key = $db->first_key();
1049 Returns the "next" key in the hash, given the previous one as the sole argument.
1050 Returns undef if there are no more keys to be fetched.
1052 $key = $db->next_key($key);
1056 Here are some examples of using hashes:
1058 my $db = DBM::Deep->new( "foo.db" );
1060 $db->put("foo", "bar");
1061 print "foo: " . $db->get("foo") . "\n";
1063 $db->put("baz", {}); # new child hash ref
1064 $db->get("baz")->put("buz", "biz");
1065 print "buz: " . $db->get("baz")->get("buz") . "\n";
1067 my $key = $db->first_key();
1069 print "$key: " . $db->get($key) . "\n";
1070 $key = $db->next_key($key);
1073 if ($db->exists("foo")) { $db->delete("foo"); }
1077 For arrays, DBM::Deep supports all the common methods described above, and the
1078 following additional methods: C<length()>, C<push()>, C<pop()>, C<shift()>,
1079 C<unshift()> and C<splice()>.
1085 Returns the number of elements in the array. Takes no arguments.
1087 my $len = $db->length();
1091 Adds one or more elements onto the end of the array. Accepts scalars, hash
1092 refs or array refs. No return value.
1094 $db->push("foo", "bar", {});
1098 Fetches the last element in the array, and deletes it. Takes no arguments.
1099 Returns undef if array is empty. Returns the element value.
1101 my $elem = $db->pop();
1105 Fetches the first element in the array, deletes it, then shifts all the
1106 remaining elements over to take up the space. Returns the element value. This
1107 method is not recommended with large arrays -- see L<LARGE ARRAYS> below for
1110 my $elem = $db->shift();
1114 Inserts one or more elements onto the beginning of the array, shifting all
1115 existing elements over to make room. Accepts scalars, hash refs or array refs.
1116 No return value. This method is not recommended with large arrays -- see
1117 <LARGE ARRAYS> below for details.
1119 $db->unshift("foo", "bar", {});
1123 Performs exactly like Perl's built-in function of the same name. See L<perldoc
1124 -f splice> for usage -- it is too complicated to document here. This method is
1125 not recommended with large arrays -- see L<LARGE ARRAYS> below for details.
1129 Here are some examples of using arrays:
1131 my $db = DBM::Deep->new(
1133 type => DBM::Deep->TYPE_ARRAY
1136 $db->push("bar", "baz");
1137 $db->unshift("foo");
1140 my $len = $db->length();
1141 print "length: $len\n"; # 4
1143 for (my $k=0; $k<$len; $k++) {
1144 print "$k: " . $db->get($k) . "\n";
1147 $db->splice(1, 2, "biz", "baf");
1149 while (my $elem = shift @$db) {
1150 print "shifted: $elem\n";
1155 Enable automatic file locking by passing a true value to the C<locking>
1156 parameter when constructing your DBM::Deep object (see L<SETUP> above).
1158 my $db = DBM::Deep->new(
1163 This causes DBM::Deep to C<flock()> the underlying filehandle with exclusive
1164 mode for writes, and shared mode for reads. This is required if you have
1165 multiple processes accessing the same database file, to avoid file corruption.
1166 Please note that C<flock()> does NOT work for files over NFS. See L<DB OVER
1167 NFS> below for more.
1169 =head2 EXPLICIT LOCKING
1171 You can explicitly lock a database, so it remains locked for multiple
1172 transactions. This is done by calling the C<lock()> method, and passing an
1173 optional lock mode argument (defaults to exclusive mode). This is particularly
1174 useful for things like counters, where the current value needs to be fetched,
1175 then incremented, then stored again.
1178 my $counter = $db->get("counter");
1180 $db->put("counter", $counter);
1189 You can pass C<lock()> an optional argument, which specifies which mode to use
1190 (exclusive or shared). Use one of these two constants: C<DBM::Deep-E<gt>LOCK_EX>
1191 or C<DBM::Deep-E<gt>LOCK_SH>. These are passed directly to C<flock()>, and are the
1192 same as the constants defined in Perl's C<Fcntl> module.
1194 $db->lock( DBM::Deep->LOCK_SH );
1198 =head1 IMPORTING/EXPORTING
1200 You can import existing complex structures by calling the C<import()> method,
1201 and export an entire database into an in-memory structure using the C<export()>
1202 method. Both are examined here.
1206 Say you have an existing hash with nested hashes/arrays inside it. Instead of
1207 walking the structure and adding keys/elements to the database as you go,
1208 simply pass a reference to the C<import()> method. This recursively adds
1209 everything to an existing DBM::Deep object for you. Here is an example:
1214 array1 => [ "elem0", "elem1", "elem2" ],
1216 subkey1 => "subvalue1",
1217 subkey2 => "subvalue2"
1221 my $db = DBM::Deep->new( "foo.db" );
1222 $db->import( $struct );
1224 print $db->{key1} . "\n"; # prints "value1"
1226 This recursively imports the entire C<$struct> object into C<$db>, including
1227 all nested hashes and arrays. If the DBM::Deep object contains exsiting data,
1228 keys are merged with the existing ones, replacing if they already exist.
1229 The C<import()> method can be called on any database level (not just the base
1230 level), and works with both hash and array DB types.
1232 B<Note:> Make sure your existing structure has no circular references in it.
1233 These will cause an infinite loop when importing.
1237 Calling the C<export()> method on an existing DBM::Deep object will return
1238 a reference to a new in-memory copy of the database. The export is done
1239 recursively, so all nested hashes/arrays are all exported to standard Perl
1240 objects. Here is an example:
1242 my $db = DBM::Deep->new( "foo.db" );
1244 $db->{key1} = "value1";
1245 $db->{key2} = "value2";
1247 $db->{hash1}->{subkey1} = "subvalue1";
1248 $db->{hash1}->{subkey2} = "subvalue2";
1250 my $struct = $db->export();
1252 print $struct->{key1} . "\n"; # prints "value1"
1254 This makes a complete copy of the database in memory, and returns a reference
1255 to it. The C<export()> method can be called on any database level (not just
1256 the base level), and works with both hash and array DB types. Be careful of
1257 large databases -- you can store a lot more data in a DBM::Deep object than an
1258 in-memory Perl structure.
1260 B<Note:> Make sure your database has no circular references in it.
1261 These will cause an infinite loop when exporting.
1265 DBM::Deep has a number of hooks where you can specify your own Perl function
1266 to perform filtering on incoming or outgoing data. This is a perfect
1267 way to extend the engine, and implement things like real-time compression or
1268 encryption. Filtering applies to the base DB level, and all child hashes /
1269 arrays. Filter hooks can be specified when your DBM::Deep object is first
1270 constructed, or by calling the C<set_filter()> method at any time. There are
1271 four available filter hooks, described below:
1275 =item * filter_store_key
1277 This filter is called whenever a hash key is stored. It
1278 is passed the incoming key, and expected to return a transformed key.
1280 =item * filter_store_value
1282 This filter is called whenever a hash key or array element is stored. It
1283 is passed the incoming value, and expected to return a transformed value.
1285 =item * filter_fetch_key
1287 This filter is called whenever a hash key is fetched (i.e. via
1288 C<first_key()> or C<next_key()>). It is passed the transformed key,
1289 and expected to return the plain key.
1291 =item * filter_fetch_value
1293 This filter is called whenever a hash key or array element is fetched.
1294 It is passed the transformed value, and expected to return the plain value.
1298 Here are the two ways to setup a filter hook:
1300 my $db = DBM::Deep->new(
1302 filter_store_value => \&my_filter_store,
1303 filter_fetch_value => \&my_filter_fetch
1308 $db->set_filter( "filter_store_value", \&my_filter_store );
1309 $db->set_filter( "filter_fetch_value", \&my_filter_fetch );
1311 Your filter function will be called only when dealing with SCALAR keys or
1312 values. When nested hashes and arrays are being stored/fetched, filtering
1313 is bypassed. Filters are called as static functions, passed a single SCALAR
1314 argument, and expected to return a single SCALAR value. If you want to
1315 remove a filter, set the function reference to C<undef>:
1317 $db->set_filter( "filter_store_value", undef );
1319 =head2 REAL-TIME ENCRYPTION EXAMPLE
1321 Here is a working example that uses the I<Crypt::Blowfish> module to
1322 do real-time encryption / decryption of keys & values with DBM::Deep Filters.
1323 Please visit L<http://search.cpan.org/search?module=Crypt::Blowfish> for more
1324 on I<Crypt::Blowfish>. You'll also need the I<Crypt::CBC> module.
1327 use Crypt::Blowfish;
1330 my $cipher = Crypt::CBC->new({
1331 'key' => 'my secret key',
1332 'cipher' => 'Blowfish',
1334 'regenerate_key' => 0,
1335 'padding' => 'space',
1339 my $db = DBM::Deep->new(
1340 file => "foo-encrypt.db",
1341 filter_store_key => \&my_encrypt,
1342 filter_store_value => \&my_encrypt,
1343 filter_fetch_key => \&my_decrypt,
1344 filter_fetch_value => \&my_decrypt,
1347 $db->{key1} = "value1";
1348 $db->{key2} = "value2";
1349 print "key1: " . $db->{key1} . "\n";
1350 print "key2: " . $db->{key2} . "\n";
1356 return $cipher->encrypt( $_[0] );
1359 return $cipher->decrypt( $_[0] );
1362 =head2 REAL-TIME COMPRESSION EXAMPLE
1364 Here is a working example that uses the I<Compress::Zlib> module to do real-time
1365 compression / decompression of keys & values with DBM::Deep Filters.
1366 Please visit L<http://search.cpan.org/search?module=Compress::Zlib> for
1367 more on I<Compress::Zlib>.
1372 my $db = DBM::Deep->new(
1373 file => "foo-compress.db",
1374 filter_store_key => \&my_compress,
1375 filter_store_value => \&my_compress,
1376 filter_fetch_key => \&my_decompress,
1377 filter_fetch_value => \&my_decompress,
1380 $db->{key1} = "value1";
1381 $db->{key2} = "value2";
1382 print "key1: " . $db->{key1} . "\n";
1383 print "key2: " . $db->{key2} . "\n";
1389 return Compress::Zlib::memGzip( $_[0] ) ;
1392 return Compress::Zlib::memGunzip( $_[0] ) ;
1395 B<Note:> Filtering of keys only applies to hashes. Array "keys" are
1396 actually numerical index numbers, and are not filtered.
1398 =head1 ERROR HANDLING
1400 Most DBM::Deep methods return a true value for success, and call die() on
1401 failure. You can wrap calls in an eval block to catch the die.
1403 my $db = DBM::Deep->new( "foo.db" ); # create hash
1404 eval { $db->push("foo"); }; # ILLEGAL -- push is array-only call
1406 print $@; # prints error message
1408 =head1 LARGEFILE SUPPORT
1410 If you have a 64-bit system, and your Perl is compiled with both LARGEFILE
1411 and 64-bit support, you I<may> be able to create databases larger than 2 GB.
1412 DBM::Deep by default uses 32-bit file offset tags, but these can be changed
1413 by calling the static C<set_pack()> method before you do anything else.
1415 DBM::Deep::set_pack(8, 'Q');
1417 This tells DBM::Deep to pack all file offsets with 8-byte (64-bit) quad words
1418 instead of 32-bit longs. After setting these values your DB files have a
1419 theoretical maximum size of 16 XB (exabytes).
1421 B<Note:> Changing these values will B<NOT> work for existing database files.
1422 Only change this for new files, and make sure it stays set consistently
1423 throughout the file's life. If you do set these values, you can no longer
1424 access 32-bit DB files. You can, however, call C<set_pack(4, 'N')> to change
1425 back to 32-bit mode.
1427 B<Note:> I have not personally tested files > 2 GB -- all my systems have
1428 only a 32-bit Perl. However, I have received user reports that this does
1431 =head1 LOW-LEVEL ACCESS
1433 If you require low-level access to the underlying filehandle that DBM::Deep uses,
1434 you can call the C<_fh()> method, which returns the handle:
1436 my $fh = $db->_fh();
1438 This method can be called on the root level of the datbase, or any child
1439 hashes or arrays. All levels share a I<root> structure, which contains things
1440 like the filehandle, a reference counter, and all the options specified
1441 when you created the object. You can get access to this root structure by
1442 calling the C<root()> method.
1444 my $root = $db->_root();
1446 This is useful for changing options after the object has already been created,
1447 such as enabling/disabling locking. You can also store your own temporary user
1448 data in this structure (be wary of name collision), which is then accessible from
1449 any child hash or array.
1451 =head1 CUSTOM DIGEST ALGORITHM
1453 DBM::Deep by default uses the I<Message Digest 5> (MD5) algorithm for hashing
1454 keys. However you can override this, and use another algorithm (such as SHA-256)
1455 or even write your own. But please note that DBM::Deep currently expects zero
1456 collisions, so your algorithm has to be I<perfect>, so to speak.
1457 Collision detection may be introduced in a later version.
1461 You can specify a custom digest algorithm by calling the static C<set_digest()>
1462 function, passing a reference to a subroutine, and the length of the algorithm's
1463 hashes (in bytes). This is a global static function, which affects ALL DBM::Deep
1464 objects. Here is a working example that uses a 256-bit hash from the
1465 I<Digest::SHA256> module. Please see
1466 L<http://search.cpan.org/search?module=Digest::SHA256> for more.
1471 my $context = Digest::SHA256::new(256);
1473 DBM::Deep::set_digest( \&my_digest, 32 );
1475 my $db = DBM::Deep->new( "foo-sha.db" );
1477 $db->{key1} = "value1";
1478 $db->{key2} = "value2";
1479 print "key1: " . $db->{key1} . "\n";
1480 print "key2: " . $db->{key2} . "\n";
1486 return substr( $context->hash($_[0]), 0, 32 );
1489 B<Note:> Your returned digest strings must be B<EXACTLY> the number
1490 of bytes you specify in the C<set_digest()> function (in this case 32).
1492 =head1 CIRCULAR REFERENCES
1494 DBM::Deep has B<experimental> support for circular references. Meaning you
1495 can have a nested hash key or array element that points to a parent object.
1496 This relationship is stored in the DB file, and is preserved between sessions.
1499 my $db = DBM::Deep->new( "foo.db" );
1502 $db->{circle} = $db; # ref to self
1504 print $db->{foo} . "\n"; # prints "foo"
1505 print $db->{circle}->{foo} . "\n"; # prints "foo" again
1507 B<Note>: Passing the object to a function that recursively walks the
1508 object tree (such as I<Data::Dumper> or even the built-in C<optimize()> or
1509 C<export()> methods) will result in an infinite loop. This will be fixed in
1512 =head1 CAVEATS / ISSUES / BUGS
1514 This section describes all the known issues with DBM::Deep. It you have found
1515 something that is not listed here, please send e-mail to L<jhuckaby@cpan.org>.
1517 =head2 UNUSED SPACE RECOVERY
1519 One major caveat with DBM::Deep is that space occupied by existing keys and
1520 values is not recovered when they are deleted. Meaning if you keep deleting
1521 and adding new keys, your file will continuously grow. I am working on this,
1522 but in the meantime you can call the built-in C<optimize()> method from time to
1523 time (perhaps in a crontab or something) to recover all your unused space.
1525 $db->optimize(); # returns true on success
1527 This rebuilds the ENTIRE database into a new file, then moves it on top of
1528 the original. The new file will have no unused space, thus it will take up as
1529 little disk space as possible. Please note that this operation can take
1530 a long time for large files, and you need enough disk space to temporarily hold
1531 2 copies of your DB file. The temporary file is created in the same directory
1532 as the original, named with a ".tmp" extension, and is deleted when the
1533 operation completes. Oh, and if locking is enabled, the DB is automatically
1534 locked for the entire duration of the copy.
1536 B<WARNING:> Only call optimize() on the top-level node of the database, and
1537 make sure there are no child references lying around. DBM::Deep keeps a reference
1538 counter, and if it is greater than 1, optimize() will abort and return undef.
1540 =head2 AUTOVIVIFICATION
1542 Unfortunately, autovivification doesn't work with tied hashes. This appears to
1543 be a bug in Perl's tie() system, as I<Jakob Schmidt> encountered the very same
1544 issue with his I<DWH_FIle> module (see L<http://search.cpan.org/search?module=DWH_File>),
1545 and it is also mentioned in the BUGS section for the I<MLDBM> module <see
1546 L<http://search.cpan.org/search?module=MLDBM>). Basically, on a new db file,
1549 $db->{foo}->{bar} = "hello";
1551 Since "foo" doesn't exist, you cannot add "bar" to it. You end up with "foo"
1552 being an empty hash. Try this instead, which works fine:
1554 $db->{foo} = { bar => "hello" };
1556 As of Perl 5.8.7, this bug still exists. I have walked very carefully through
1557 the execution path, and Perl indeed passes an empty hash to the STORE() method.
1558 Probably a bug in Perl.
1562 (The reasons given assume a high level of Perl understanding, specifically of
1563 references. You can safely skip this section.)
1565 Currently, the only references supported are HASH and ARRAY. The other reference
1566 types (SCALAR, CODE, GLOB, and REF) cannot be supported for various reasons.
1572 These are things like filehandles and other sockets. They can't be supported
1573 because it's completely unclear how DBM::Deep should serialize them.
1575 =item * SCALAR / REF
1577 The discussion here refers to the following type of example:
1584 # In some other process ...
1586 my $val = ${ $db->{key1} };
1588 is( $val, 50, "What actually gets stored in the DB file?" );
1590 The problem is one of synchronization. When the variable being referred to
1591 changes value, the reference isn't notified. This means that the new value won't
1592 be stored in the datafile for other processes to read. There is no TIEREF.
1594 It is theoretically possible to store references to values already within a
1595 DBM::Deep object because everything already is synchronized, but the change to
1596 the internals would be quite large. Specifically, DBM::Deep would have to tie
1597 every single value that is stored. This would bloat the RAM footprint of
1598 DBM::Deep at least twofold (if not more) and be a significant performance drain,
1599 all to support a feature that has never been requested.
1603 L<http://search.cpan.org/search?module=Data::Dump::Streamer> provides a
1604 mechanism for serializing coderefs, including saving off all closure state.
1605 However, just as for SCALAR and REF, that closure state may change without
1606 notifying the DBM::Deep object storing the reference.
1610 =head2 FILE CORRUPTION
1612 The current level of error handling in DBM::Deep is minimal. Files I<are> checked
1613 for a 32-bit signature when opened, but other corruption in files can cause
1614 segmentation faults. DBM::Deep may try to seek() past the end of a file, or get
1615 stuck in an infinite loop depending on the level of corruption. File write
1616 operations are not checked for failure (for speed), so if you happen to run
1617 out of disk space, DBM::Deep will probably fail in a bad way. These things will
1618 be addressed in a later version of DBM::Deep.
1622 Beware of using DB files over NFS. DBM::Deep uses flock(), which works well on local
1623 filesystems, but will NOT protect you from file corruption over NFS. I've heard
1624 about setting up your NFS server with a locking daemon, then using lockf() to
1625 lock your files, but your mileage may vary there as well. From what I
1626 understand, there is no real way to do it. However, if you need access to the
1627 underlying filehandle in DBM::Deep for using some other kind of locking scheme like
1628 lockf(), see the L<LOW-LEVEL ACCESS> section above.
1630 =head2 COPYING OBJECTS
1632 Beware of copying tied objects in Perl. Very strange things can happen.
1633 Instead, use DBM::Deep's C<clone()> method which safely copies the object and
1634 returns a new, blessed, tied hash or array to the same level in the DB.
1636 my $copy = $db->clone();
1638 B<Note>: Since clone() here is cloning the object, not the database location, any
1639 modifications to either $db or $copy will be visible in both.
1643 Beware of using C<shift()>, C<unshift()> or C<splice()> with large arrays.
1644 These functions cause every element in the array to move, which can be murder
1645 on DBM::Deep, as every element has to be fetched from disk, then stored again in
1646 a different location. This will be addressed in the forthcoming version 1.00.
1648 =head2 WRITEONLY FILES
1650 If you pass in a filehandle to new(), you may have opened it in either a readonly or
1651 writeonly mode. STORE will verify that the filehandle is writable. However, there
1652 doesn't seem to be a good way to determine if a filehandle is readable. And, if the
1653 filehandle isn't readable, it's not clear what will happen. So, don't do that.
1657 This section discusses DBM::Deep's speed and memory usage.
1661 Obviously, DBM::Deep isn't going to be as fast as some C-based DBMs, such as
1662 the almighty I<BerkeleyDB>. But it makes up for it in features like true
1663 multi-level hash/array support, and cross-platform FTPable files. Even so,
1664 DBM::Deep is still pretty fast, and the speed stays fairly consistent, even
1665 with huge databases. Here is some test data:
1667 Adding 1,000,000 keys to new DB file...
1669 At 100 keys, avg. speed is 2,703 keys/sec
1670 At 200 keys, avg. speed is 2,642 keys/sec
1671 At 300 keys, avg. speed is 2,598 keys/sec
1672 At 400 keys, avg. speed is 2,578 keys/sec
1673 At 500 keys, avg. speed is 2,722 keys/sec
1674 At 600 keys, avg. speed is 2,628 keys/sec
1675 At 700 keys, avg. speed is 2,700 keys/sec
1676 At 800 keys, avg. speed is 2,607 keys/sec
1677 At 900 keys, avg. speed is 2,190 keys/sec
1678 At 1,000 keys, avg. speed is 2,570 keys/sec
1679 At 2,000 keys, avg. speed is 2,417 keys/sec
1680 At 3,000 keys, avg. speed is 1,982 keys/sec
1681 At 4,000 keys, avg. speed is 1,568 keys/sec
1682 At 5,000 keys, avg. speed is 1,533 keys/sec
1683 At 6,000 keys, avg. speed is 1,787 keys/sec
1684 At 7,000 keys, avg. speed is 1,977 keys/sec
1685 At 8,000 keys, avg. speed is 2,028 keys/sec
1686 At 9,000 keys, avg. speed is 2,077 keys/sec
1687 At 10,000 keys, avg. speed is 2,031 keys/sec
1688 At 20,000 keys, avg. speed is 1,970 keys/sec
1689 At 30,000 keys, avg. speed is 2,050 keys/sec
1690 At 40,000 keys, avg. speed is 2,073 keys/sec
1691 At 50,000 keys, avg. speed is 1,973 keys/sec
1692 At 60,000 keys, avg. speed is 1,914 keys/sec
1693 At 70,000 keys, avg. speed is 2,091 keys/sec
1694 At 80,000 keys, avg. speed is 2,103 keys/sec
1695 At 90,000 keys, avg. speed is 1,886 keys/sec
1696 At 100,000 keys, avg. speed is 1,970 keys/sec
1697 At 200,000 keys, avg. speed is 2,053 keys/sec
1698 At 300,000 keys, avg. speed is 1,697 keys/sec
1699 At 400,000 keys, avg. speed is 1,838 keys/sec
1700 At 500,000 keys, avg. speed is 1,941 keys/sec
1701 At 600,000 keys, avg. speed is 1,930 keys/sec
1702 At 700,000 keys, avg. speed is 1,735 keys/sec
1703 At 800,000 keys, avg. speed is 1,795 keys/sec
1704 At 900,000 keys, avg. speed is 1,221 keys/sec
1705 At 1,000,000 keys, avg. speed is 1,077 keys/sec
1707 This test was performed on a PowerMac G4 1gHz running Mac OS X 10.3.2 & Perl
1708 5.8.1, with an 80GB Ultra ATA/100 HD spinning at 7200RPM. The hash keys and
1709 values were between 6 - 12 chars in length. The DB file ended up at 210MB.
1710 Run time was 12 min 3 sec.
1714 One of the great things about DBM::Deep is that it uses very little memory.
1715 Even with huge databases (1,000,000+ keys) you will not see much increased
1716 memory on your process. DBM::Deep relies solely on the filesystem for storing
1717 and fetching data. Here is output from I</usr/bin/top> before even opening a
1720 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1721 22831 root 11 0 2716 2716 1296 R 0.0 0.2 0:07 perl
1723 Basically the process is taking 2,716K of memory. And here is the same
1724 process after storing and fetching 1,000,000 keys:
1726 PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
1727 22831 root 14 0 2772 2772 1328 R 0.0 0.2 13:32 perl
1729 Notice the memory usage increased by only 56K. Test was performed on a 700mHz
1730 x86 box running Linux RedHat 7.2 & Perl 5.6.1.
1732 =head1 DB FILE FORMAT
1734 In case you were interested in the underlying DB file format, it is documented
1735 here in this section. You don't need to know this to use the module, it's just
1736 included for reference.
1740 DBM::Deep files always start with a 32-bit signature to identify the file type.
1741 This is at offset 0. The signature is "DPDB" in network byte order. This is
1742 checked for when the file is opened and an error will be thrown if it's not found.
1746 The DBM::Deep file is in a I<tagged format>, meaning each section of the file
1747 has a standard header containing the type of data, the length of data, and then
1748 the data itself. The type is a single character (1 byte), the length is a
1749 32-bit unsigned long in network byte order, and the data is, well, the data.
1750 Here is how it unfolds:
1754 Immediately after the 32-bit file signature is the I<Master Index> record.
1755 This is a standard tag header followed by 1024 bytes (in 32-bit mode) or 2048
1756 bytes (in 64-bit mode) of data. The type is I<H> for hash or I<A> for array,
1757 depending on how the DBM::Deep object was constructed.
1759 The index works by looking at a I<MD5 Hash> of the hash key (or array index
1760 number). The first 8-bit char of the MD5 signature is the offset into the
1761 index, multipled by 4 in 32-bit mode, or 8 in 64-bit mode. The value of the
1762 index element is a file offset of the next tag for the key/element in question,
1763 which is usually a I<Bucket List> tag (see below).
1765 The next tag I<could> be another index, depending on how many keys/elements
1766 exist. See L<RE-INDEXING> below for details.
1770 A I<Bucket List> is a collection of 16 MD5 hashes for keys/elements, plus
1771 file offsets to where the actual data is stored. It starts with a standard
1772 tag header, with type I<B>, and a data size of 320 bytes in 32-bit mode, or
1773 384 bytes in 64-bit mode. Each MD5 hash is stored in full (16 bytes), plus
1774 the 32-bit or 64-bit file offset for the I<Bucket> containing the actual data.
1775 When the list fills up, a I<Re-Index> operation is performed (See
1776 L<RE-INDEXING> below).
1780 A I<Bucket> is a tag containing a key/value pair (in hash mode), or a
1781 index/value pair (in array mode). It starts with a standard tag header with
1782 type I<D> for scalar data (string, binary, etc.), or it could be a nested
1783 hash (type I<H>) or array (type I<A>). The value comes just after the tag
1784 header. The size reported in the tag header is only for the value, but then,
1785 just after the value is another size (32-bit unsigned long) and then the plain
1786 key itself. Since the value is likely to be fetched more often than the plain
1787 key, I figured it would be I<slightly> faster to store the value first.
1789 If the type is I<H> (hash) or I<A> (array), the value is another I<Master Index>
1790 record for the nested structure, where the process begins all over again.
1794 After a I<Bucket List> grows to 16 records, its allocated space in the file is
1795 exhausted. Then, when another key/element comes in, the list is converted to a
1796 new index record. However, this index will look at the next char in the MD5
1797 hash, and arrange new Bucket List pointers accordingly. This process is called
1798 I<Re-Indexing>. Basically, a new index tag is created at the file EOF, and all
1799 17 (16 + new one) keys/elements are removed from the old Bucket List and
1800 inserted into the new index. Several new Bucket Lists are created in the
1801 process, as a new MD5 char from the key is being examined (it is unlikely that
1802 the keys will all share the same next char of their MD5s).
1804 Because of the way the I<MD5> algorithm works, it is impossible to tell exactly
1805 when the Bucket Lists will turn into indexes, but the first round tends to
1806 happen right around 4,000 keys. You will see a I<slight> decrease in
1807 performance here, but it picks back up pretty quick (see L<SPEED> above). Then
1808 it takes B<a lot> more keys to exhaust the next level of Bucket Lists. It's
1809 right around 900,000 keys. This process can continue nearly indefinitely --
1810 right up until the point the I<MD5> signatures start colliding with each other,
1811 and this is B<EXTREMELY> rare -- like winning the lottery 5 times in a row AND
1812 getting struck by lightning while you are walking to cash in your tickets.
1813 Theoretically, since I<MD5> hashes are 128-bit values, you I<could> have up to
1814 340,282,366,921,000,000,000,000,000,000,000,000,000 keys/elements (I believe
1815 this is 340 unodecillion, but don't quote me).
1819 When a new key/element is stored, the key (or index number) is first run through
1820 I<Digest::MD5> to get a 128-bit signature (example, in hex:
1821 b05783b0773d894396d475ced9d2f4f6). Then, the I<Master Index> record is checked
1822 for the first char of the signature (in this case I<b0>). If it does not exist,
1823 a new I<Bucket List> is created for our key (and the next 15 future keys that
1824 happen to also have I<b> as their first MD5 char). The entire MD5 is written
1825 to the I<Bucket List> along with the offset of the new I<Bucket> record (EOF at
1826 this point, unless we are replacing an existing I<Bucket>), where the actual
1827 data will be stored.
1831 Fetching an existing key/element involves getting a I<Digest::MD5> of the key
1832 (or index number), then walking along the indexes. If there are enough
1833 keys/elements in this DB level, there might be nested indexes, each linked to
1834 a particular char of the MD5. Finally, a I<Bucket List> is pointed to, which
1835 contains up to 16 full MD5 hashes. Each is checked for equality to the key in
1836 question. If we found a match, the I<Bucket> tag is loaded, where the value and
1837 plain key are stored.
1839 Fetching the plain key occurs when calling the I<first_key()> and I<next_key()>
1840 methods. In this process the indexes are walked systematically, and each key
1841 fetched in increasing MD5 order (which is why it appears random). Once the
1842 I<Bucket> is found, the value is skipped and the plain key returned instead.
1843 B<Note:> Do not count on keys being fetched as if the MD5 hashes were
1844 alphabetically sorted. This only happens on an index-level -- as soon as the
1845 I<Bucket Lists> are hit, the keys will come out in the order they went in --
1846 so it's pretty much undefined how the keys will come out -- just like Perl's
1849 =head1 CODE COVERAGE
1851 We use B<Devel::Cover> to test the code coverage of our tests, below is the
1852 B<Devel::Cover> report on this module's test suite.
1854 ----------------------------------- ------ ------ ------ ------ ------ ------
1855 File stmt bran cond sub time total
1856 ----------------------------------- ------ ------ ------ ------ ------ ------
1857 blib/lib/DBM/Deep.pm 94.9 80.6 73.0 100.0 37.9 90.4
1858 blib/lib/DBM/Deep/Array.pm 100.0 91.1 100.0 100.0 18.2 98.1
1859 blib/lib/DBM/Deep/Engine.pm 98.9 87.3 80.0 100.0 34.2 95.2
1860 blib/lib/DBM/Deep/Hash.pm 100.0 87.5 100.0 100.0 9.7 97.3
1861 Total 97.9 85.9 79.7 100.0 100.0 94.3
1862 ----------------------------------- ------ ------ ------ ------ ------ ------
1864 =head1 MORE INFORMATION
1866 Check out the DBM::Deep Google Group at L<http://groups.google.com/group/DBM-Deep>
1867 or send email to L<DBM-Deep@googlegroups.com>.
1871 Joseph Huckaby, L<jhuckaby@cpan.org>
1873 Rob Kinyon, L<rkinyon@cpan.org>
1875 Special thanks to Adam Sah and Rich Gaushell! You know why :-)
1879 perltie(1), Tie::Hash(3), Digest::MD5(3), Fcntl(3), flock(2), lockf(3), nfs(5),
1880 Digest::SHA256(3), Crypt::Blowfish(3), Compress::Zlib(3)
1884 Copyright (c) 2002-2006 Joseph Huckaby. All Rights Reserved.
1885 This is free software, you may use it and distribute it under the
1886 same terms as Perl itself.